Statistical Analysis Of Bayesian Transformation Model Of Complex Data

In survival analysis,it is worth paying attention to study the remaining life of patients and the factors affecting the survival time of patients.In many disciplines,such as medicine,biology and so on,researchers are interested in the time when something happens,but the data generated by practical problems are often incomplete.Because researchers are interested in the development of events after a certain time or the life expectancy of patients in different medical institutions.So the event time we are concerned about may have left truncation right deletion or grouping interval deletion.Therefore,this paper will give the related research on left truncated right censored data and grouped interval I censored data.The first part of this paper mainly discusses the estimation of Box-Cox transformation model under left truncated and right censored data.Firstly,the data types and research status of left truncated and right censored data are introduced;Secondly,according to Box-Cox transformation,the corresponding model is established,and a piecewise constant risk model is established for the baseline hazards function.Then the corresponding posterior likelihood function is deduced by given truncated normal prior;Finally,Bayesian estimation method is used to estimate the model parameters.In the process of Bayesian estimation,since the posterior likelihood of the parameters to be estimated does not have the standard distribution form,the ARMS algorithm in MCMC sampling algorithm is used for sampling estimation.The simulation results verify the effectiveness of this method.Finally,the estimation method is applied to the data of a retirement center for example analysis.The second part of this paper mainly discusses the estimation of Box-Cox transformation model with cluster interval censored data.Firstly,the characteristics and research status of cluster failure time data and current status data are introduced;Secondly,in order to better explain the correlation in groups,random effects are added to the Box-Cox transformation model.And a piecewise constant risk model is established for the baseline hazards function.Subsequently,the likelihood function under grouped interval type I censored data is constructed.Then the posterior likelihood function is deduced by given truncated normal prior.In the process of Bayesian estimation,since the posterior likelihood of the parameters to be estimated does not have the standard distribution form,the ARMS algorithm in MCMC sampling algorithm is used for sampling estimation.The validity of Bayesian transformation model is verified by simulation.